What causes greater deviations from predictions of metabolic scaling theory in earlier successional forests?

- Scaling relationships conform to Metabolic scaling theory (MST) in primary forests.
- But earlier successional forests reveal greater deviations from MST predictions.
- Diameter-height scaling is mainly affected by competition and successional stage.
- Tree-size distribution scaling is related to recruitment limitation and succession.
- Both scalings are not affected by size, and were indirectly influenced by climate.

Metabolic scaling theory (MST) is frequently criticized because its predictions do not match many observations that allometric relationships change with climate and species. Understanding why observations deviate from theory predictions is critical for better forest carbon accounting and management in a rapid warming world. Here we used forest plots across successional and latitudinal gradients to test the predictions that: (1) observed exponents should deviate more from MST predictions at earlier successional stages; (2) the deviations may be related to changes in competition, recruitment limitation and tree size across successional stages and climate gradient. We sampled forest plots (each 1000 m2) from four successional stages (early to late) in four sites along a latitudinal gradient (42-50°N) in Northeast China, and examined the scaling relationship between tree diameter and height (D-H) and plot-level tree size distribution (D-N scaling). We related the scaling exponents for each plot to successional stage, climate, local topographic factors, and proxies for tree size, light competition and recruitment limitation. We used mixed-effect structural equation model to examine the major factors causing the deviations from MST predictions. The results showed that both D-H and D-N scaling exponents conformed to MST predictions in late-successional forests, but D-H exponent deceased while D-N exponent increased regularly towards later-successional stages. Both exponents were little correlated with climate and local topographic factors, but were significantly correlated to tree size, and proxies for competition (stem density) and recruitment limitation (skewness of tree size distribution). D-H scaling was mainly affected by stem density and successional stage, while D-N scaling mainly affected by skewness and succession. Both scaling relationships were not significantly affected by tree size, and were indirectly influenced by climate through stem density or skewness. Our results provide clear evidence that MST predictions are only supported in the late successional forests, while forests at an earlier successional stage reveal greater deviations. These deviations may be caused by processes not considered in MST, including light competition, demography dynamics and successional status. Future MST need to incorporate these ecological processes to better predict the allometries for disturbed forests, which are widespread across the world and have important roles in carbon sequestration.